Efficacy of intravesical electrical stimulation and extracorporeal shock wave therapy for female non-obstructive detrusor underactivity: a reinforcement learning optimization approach
10.3969/j.issn.1009-8291.2025.12.007
- VernacularTitle:引入强化学习优化策略的膀胱腔内电刺激与冲击波治疗女性非梗阻性逼尿肌活动低下的疗效
- Author:
Zhen WANG
1
;
Zhongle XU
2
;
Junhua XI
2
;
Yanbin ZHANG
1
Author Information
1. Wannan Medical College, Wuhu 241002; Department of Urology, The Second People's Hospital of Hefei, Hefei 230011, China
2. Department of Urology, The Second People's Hospital of Hefei, Hefei 230011, China
- Publication Type:Journal Article
- Keywords:
detrusor underactivity;
intravesical electrical stimulation;
low-intensity extracorporeal shock wave therapy;
reinforcement learning;
artificial intelligence;
voiding dysfunction;
individualized comprehensive treatment
- From:
Journal of Modern Urology
2025;30(12):1056-1063
- CountryChina
- Language:Chinese
-
Abstract:
Objective To evaluate the efficacy of intravesical electrical stimulation (IVES) and low-intensity extracorporeal shock wave therapy (Li-ESWT) in improving bladder emptying function in female patients with non-obstructive detrusor underactivity (NODU), and to further assess the clinical value of an individualized integrated treatment strategy guided by reinforcement learning (RL) optimization. Methods A total of 98 female patients diagnosed with NODU by urodynamic testing at the Department of Urology, the Second People's Hospital of Hefei, duirng Jun.2023 and Feb.2025 were prospectively enrolled. Patients were randomly assigned (1∶1∶1) to three groups:the IVES group (n=33), the Li-ESWT group (n=35), and the RL group (n=30). Clinical outcomes before and after the 4-week treatment were compared among the three groups, including peak detrusor pressure during urination (PdetQmax), maximum urinary flow rate (Qmax), post-void residual (PVR), bladder contractility index (BCI), patient perception of bladder condition-scale (PPBC-S), incontinence impact questionnaire-short form 7 (IIQ-7), urogenital distress inventory-short form 6 (UDI-6), total efficiency and satisfaction.A RL model was trained based on clinical data, with a model structure diagram and reward convergence curve plotted to validate the utility of the RL system in optimizing individualized treatment parameters. Results There were no statistically significant differences in baseline characteristics among the three groups (P>0.05). After 4 weeks of treatment, all groups demonstrated significant improvements in PdetQmax, Qmax, and BCI, along with significant reductions in PVR, PPBC-S, IIQ-7, and UDI-6 scores (all P<0.01). Notably, the RL group exhibited significantly greater improvements in PdetQmax, Qmax, and BCI, and more pronounced reductions in PVR, PPBC-S, IIQ-7, and UDI-6 scores than the IVES and Li-ESWT groups (all P<0.05). Specifically, the RL group showed the most substantial improvements in Qmax, PVR, and BCI than the other two groups (all P<0.01). The total effective rate in the RL group was 90.0% (27/30), which was higher than that of the IVES group (81.8%,27/33) and the Li-ESWT group (77.1%,27/35), but the differences were not statistically significant (χ
=2.63, P=0.27). The Li-ESWT group had a satisfaction rate of 51.4% (18/35), which was higher than that of the RL group (30.0%,9/30) and the IVES group (27.3%,9/33), but the differences were not statistically significant (χ
=6.76, P=0.34). No serious adverse events were observed in any group. After approximately 200 iterations, the reward value of the RL agent stabilized, and the individualized treatment parameters recommended further optimized bladder emptying efficiency. Conclusion Compared to IVES and Li-ESWT, the RL-optimized individualized comprehensive treatment strategy can significantly improve the bladder emptying function in women with NODU.
